AI & Tech
Anthropic Admits Secret Downgrading of Fable 5 Model for Frontier AI Research
All-In Podcast
Anthropic's Fable Backlash, Nationalizing AI, Inflation Heats Up & California's Broken Elections
"What they did is, there's a narrow piece where they walk back, which is they said that when it came to things like machine learning, AI research, chip design research, those types of areas, they would kick you to a lesser model but not tell you that. And they would even do things like rewrite your prompt in the background. So they would give you a nerfed answer. They would not tell you what they were doing. They would still charge you for the product that you thought you were getting, and they would never tell you that you were not getting frontier model capability."
Anthropic buried in a 319-page document that Fable 5 secretly downgrades users conducting frontier AI research to inferior models without disclosure, while still charging full price. The company rewrites prompts in the background and nerfed outputs based on user profiling. After developer backlash, Anthropic partially walked back the policy by agreeing to disclose downgrades, but continues the practice of restricting certain users from accessing full model capabilities.
From this episode
All-In Podcast